When evaluating the literature on the treatment of late-life depression, it is most critical to consider the methodology of each study, specifically: (1) Is the antidepressant treatment adequate with respect to dosing duration? (2) How is response defined, e.g., is it simply a 50% reduction in a scale score, or are there criteria for establishing which patients have truly recovered? (3) Are the outcome data given for both the intent-to-treat and completer analyses? A review of studies that meet a rigorous standard of methodology shows that (1) a therapeutic plasma nortriptyline level consistently produces a 70% to 80% remission rate in depressed patients over 60 years of age, (2) there is some scatter in the remission rates reported for fluoxetine with results ranging from 21% to 50%, and (3) studies of sertraline consistently report a remission rate of 50% or higher. Contrary to the widely held clinical belief, tricyclic treatment is not associated with a higher dropout rate compared with treatment with a serotonin selective reuptake inhibitor (SSRI). However, patients who recovered using tricyclics have lower scores on health-related quality-of-life scales than patients who recovered using SSRIs, and the long-term impact of the "tolerated" side effects of tricyclics, specifically, increased heart rate and anticholinergic effects, may be deleterious.